Skip to main content
. 2021 Feb 17;21:367. doi: 10.1186/s12889-021-10356-0

Table 3.

Results of the OLS regression showing β coefficients adjusted for covariates for recall scores, sentiment scores, and willingness to use a mask for each infographic group

Outcome Measures
Recall
Score
Sentiment
Score
Willingness
to Use a Mask
β 95% CI β 95% CI β 95% CI
UK Gov −0.061 −0.212, 0.090 Ref Ref −0.037* --0.070, −0.004
EU CDC 0.140 −0.013, 0.294 0.120 −0.014, 0.255 −0.022 −0.056, 0.011
WHO 0.436*** 0.277, 0.596 0.168* 0.029, 0.307 0.008 −0.026, 0.043
SMOH 0.432*** 0.279, 0.584 0.333*** 0.199, 0.467 −0.016 − 0.049, 0.017
BIT 0.400*** 0.245, 0.555 0.315*** 0.179, 0.450 −0.012 −0.046, 0.021
Gender: Male −0.461*** − 0.556, − 0.366 −0.072 − 0.159, 0.016 −0.020 − 0.041, 0.0003
Age Group: 40–59 0.496*** 0.386, 0.606 0.156** 0.055, 0.258 −0.041*** − 0.065, − 0.017
Age Group: 60+ 0.710*** 0.585, 0.836 0.225*** 0.110, 0.341 −0.024 − 0.051, 0.003
Ethnicity: Asian − 0.754*** − 0.943, − 0.565 0.106 − 0.069, 0.280 0.070*** 0.029, 0.111
Ethnicity: Black −0.529*** − 0.810, − 0.249 0.182 − 0.082, 0.446 0.043 − 0.018, 0.104
Ethnicity: Mixed/other −0.577*** − 0.816, − 0.337 −0.182 − 0.400, 0.036 0.006 − 0.046, 0.058
Education: No degree −0.022 − 0.131, 0.087 0.005 − 0.095, 0.105 −0.013 − 0.037, 0.011
Constant 7.010*** 6.860, 7.150 3.680*** 3.540, 3.820 0.914*** 0.882, 0.946
Observations 4099 4099 3090 3090 4099 4099
R2 0.015 0.099 0.010 0.019 0.002 0.010
Adjusted R2 0.013 0.096 0.009 0.015 0.001 0.007
Residual Std. Error 1.610 1.540 1.240 1.230 0.337 0.336
F Statistic 12.200*** 34.500*** 8.070*** 4.920*** 1.500 3.210***

*p**p***p < 0.001

Note 1: The reference group for recall score and willingness to use a mask was the no-stimuli control condition. The reference group for sentiment score was UK government’s text guidance, thus the number of observations was smaller than that of the other outcome measures. The reference age group was 18–39 years old, the reference ethnicity group is White, and the reference education group is with degree

Note 2: Adjusting for covariates did not affect the sign and significance level compared to non-adjusted analysis (not shown)